The present invention relates to digital modulation/demodulation used for satellite communication, mobile communication, mobile satellite communication, in particular, to a demodulator periodically inserting a known signal into transmitted signals, estimating channel distortion from the known signal, eliminating the channel distortion using the estimated value, and performing coherent detection of the received signal.

Recently, in the field of satellite communication, mobile communication, and mobile satellite communication, an investigation of digital modulation/demodulation has been actively performed. Especially, in environments of the mobile communication, a signal is received with fading. Various demodulation systems are examined, which stably operates under such fading environment. Among these systems, a system is remarkably noticed as a system capable to perform absolute coherent detection under fading environment, in which a known signal is periodically inserted into transmitted signals for calibration of channel distortion such as fading, and estimation and compensation of fading distortion is performed based on the known signal.

In the accompanying drawings Figs. 15 and 16 show conventional demodulation circuits disclosed in, for example, "Rayleigh Fading Compensation Method for 16QAM MODEM in Digital Land Mobile Radio Systems" (Sampei, Theses of The Institute of Electronics, Information and Communication Engineers (B-II), J72-II, No. 1, pp7 - 15 (1989-1)). In Fig. 15, a reference numeral 40 shows a quasi-coherent detecting unit, a reference numeral 50p shows a carrier estimator and 60 shows a data discriminator. In Fig. 23, a reference numeral 51 shows a fading distortion estimator and 52 shows a fading distortion compensator.

In the following, an operation of the above demodulation circuit will be explained.

Here, it is assumed a signal is modulated by QPSK (Quadriphase Phase Shift Keying). A transmitted signal S_T(t), modulated by QPSK, is given by the following expression (1).

In expression (1), Re[·] shows a real part of [·] and f_c shows a carrier frequency. z_T(t) shows a transmitted baseband signal and is given by expression (2).$${\mathit{\text{s}}}_{\mathit{\text{T}}}\text{(}\mathit{\text{t}}\text{) = Re[}{\mathit{\text{z}}}_{\mathit{\text{T}}}\text{(}\mathit{\text{t}}\text{)exp(}\mathit{\text{j}}\text{2π}{\mathit{\text{f}}}_{\mathit{\text{c}}}\mathit{\text{t}}\text{)]}$$$${\mathit{\text{z}}}_{\mathit{\text{T}}}\text{(}\mathit{\text{t}}\text{) =}{\mathit{\text{z}}}_{\mathit{\text{p}}}\text{(}\mathit{\text{t}}\text{) +}\mathit{\text{j}}\text{·}{\mathit{\text{z}}}_{\mathit{\text{Q}}}\text{(}\mathit{\text{t}}\text{)}$$

In case of the above signal transmitted under the fading environment, the faded signal is represented by a narrow-band random complex signal c(t). c(t) is calculated by multiplying fading power spectrum S(f) by S_T(t). The received signal S_R(t) faded by an envelope variation c(t) is given by expression (3).$${\mathit{\text{s}}}_{\mathit{\text{R}}}\text{(}\mathit{\text{t}}\text{) = Re[}\mathit{\text{c}}\text{(}\mathit{\text{t}}\text{)}{\mathit{\text{z}}}_{\mathit{\text{R}}}\text{(}\mathit{\text{t}}\text{)exp(}\mathit{\text{j}}\text{2}{\mathit{\text{πf}}}_{\mathit{\text{c}}}\mathit{\text{t}}\text{) +}\mathit{\text{n}}\text{(}\mathit{\text{t}}\text{) exp(}\mathit{\text{j}}\text{2π}{\mathit{\text{f}}}_{\mathit{\text{c}}}\mathit{\text{t}}\text{)]}$$

In the above expression (3), z_R(t) denotes the received baseband complex signal limited by the transmitter and receiver filters. n(t) denotes a white Gaussian noise.

In the following, the receiver will be discussed.

The quasi-coherent detection is performed on the received signal by the quasi-coherent detecting unit 40 with a local oscillator having oscillating frequency of f_c - f_off. A signal S_RB(t) output from the quasi-coherent detecting unit 40 is given by the following expression (4).$$\mathit{\text{u}}\text{(}\mathit{\text{t}}\text{) =}\mathit{\text{c}}\text{(}\mathit{\text{t}}\text{)}{\mathit{\text{z}}}_{\mathit{\text{R}}}\text{(}\mathit{\text{t}}\text{) +}\mathit{\text{n}}\text{(}\mathit{\text{t}}\text{)}$$

The carrier estimator 50p obtains c_R(t), the estimation value of c(t) from the baseband signal u(t) output from the quasi-coherent detecting unit 40. The carrier estimator 50p also eliminates the distortion caused by c(t) from the received signal. Here, it is assumed that the frequency offset value f_off is sufficiently small (f_off = 0). When the known signal is supposed to be a_p, z_R(t) becomes z_R(t) = a_p. Therefore, the received baseband complex signal u(t) is given by the following expression (5).

Accordingly, c_R(t), the estimation value of c(t) is obtained by the expression (6).$$\mathit{\text{u}}\text{(}\mathit{\text{t}}\text{) =}\mathit{\text{c}}\text{(}\mathit{\text{t}}\text{)}{\mathit{\text{a}}}_{\mathit{\text{p}}}\text{+}\mathit{\text{n}}\text{(}\mathit{\text{t}}\text{)}$$

Based on c_R(t) estimated for each known signal, the estimation sequence c_RS(t) is represented by the following expression (7) when an insertion interval of the known signal is T_R.

An estimation of an interpolation sequence c_RI(t) out of c_RS(t) is, for example, performed by the following way using the interpolation.

In the estimation of fading distortion using the interpolation, the fading distortion of the information signal among the known signal is estimated by the interpolation using the estimation value of the fading distortion obtained by the known signal. Namely, in the fading distortion estimator 51 estimates the fading distortion c_R(k-1), c_R(k), c_R(k+1), obtained at t = (k-1)T_R, kT_R, (k+1)T_R corresponding to the known signal among the signals output from the quasi-coherent detecting unit 40 as shown in Fig. 17 The fading distortion at t = kT_R + (m/N_R) T_R (N_R: the number of symbols corresponding to the insertion interval of the known signal) is estimated, for example, using the second order Gaussian interpolation and given by the following expression (8).

The fading compensator 52 eliminates the fading distortion from the received signal using the interpolation sequence c_RI(k+m/N_R) output from the fading distortion estimator 51. Namely, a transmission function h(k+m/N_R) for compensating the fading distortion is represented by the following expression (9).

From the expression (9), the fading distortion of the signal output from the quasi-coherent detecting unit 40 is compensated as shown in the expression (10).

As described above, the received signal z_RE(k+m/N_R), of which fading distortion was compensated, can be obtained.

According to the conventional fading distortion estimation and compensation method, the fading distortion of the information signal is estimated by the interpolation using the estimation value of the fading distortion obtained by the known signal. Therefore, in the nonfading channel or Rician fading channel, where thermal noise influences a lot in addition to the fading variation, an error of the estimation value obtained by the known signal becomes large, the fading distortion cannot be compensated properly, which degrades performance of the receiver such as a bit error rate characteristic.

Further, even in the fading channel, thermal noise influences a lot when C/N (carrier power to noise power rate) is low, and the estimation error of the fading distortion estimated by the known signal becomes large. The fading distortion cannot be compensated properly, which degrades the performance of the receiver such as the bit error rate characteristic.

Further, the fading channel and the nonfading channel have opposite characteristics, so that when the performance of the receiver is improved in one of two above channels, the performance of the receiver is degraded in the other channel. It is difficult to embody the receiver having a high performance in wide range from the fading channel to nonfading channel.

US 5,875,215, upon which the precharacterising portions of appended claims 1 and 4 are based, describes a carrier synchronising unit in which interpolation and recursive least square type phase and amplitude estimation of pilot symbols are used to estimate fading channel distortion.

The present invention is provided to solve the abovementioned problems. The invention aims to have a demodulator, where the estimation error of the fading distortion is reduced to improve the receiving performance such as the bit error rate characteristic not only in the fading channel, but also in a channel where the thermal noise influences a lot in addition to the fading variation such as the nonfading channel and Rician fading channel, or, in the fading channel where C/N rate is low, the estimation error of the fading distortion is made small to improve the receiving performance such as the bit error rate characteristic.

According to the present invention, a demodulator for estimating a carrier, and for coherent detection using specific symbols inserted in received signals, having a fading distortion estimator and a fading distortion compensator using an output of the fading distortion estimator, the demodulator comprising a signal compensator for compensating the received signals using a mean value characterised by further comprising a modulation component eliminator for filtering the received signals outputted from the fading distortion compensator, to produce estimated-received signals; and an averaging unit for averaging the estimated-received signals to produce said mean, or average, value wherein the fading distortion estimator has a distortion detecting/averaging unit for having an average distortion value of the specific symbols, and has a weighted averaging unit for accumulating the average distortion value after multiplying with weight factors.

According to the invention, a demodulation method in a demodulator having a fading distortion estimator and a fading distortion compensator using an output of the fading distortion estimator, comprising a signal compensation step for compensating the received signals using a mean value of the estimated-received signals, characterised by further comprising: a modulation component eliminating step for filtering received signals including specific symbols from the fading distortion estimator, to produce estimated-received signals; an averaging step for obtaining said mean value of the estimated-received signals; a distortion detecting averaging step for obtaining an average distortion value of specific symbols; and a weighted averaging step for accumulating the average distortion value after multiplying with weight factors.

The present invention will be further described by way of example with reference to the accompanying drawings, wherein:

• Fig. 1 shows a configuration of a transmitter according to the first embodiment of the present invention;
• Fig. 2 shows a configuration of the transmitted signal according to the first embodiment of the present invention;
• Fig. 3 shows a configuration of a receiver according to the first embodiment;
• Fig. 4 shows a configuration of a carrier estimator according to the first embodiment;
• Fig. 5 shows a configuration of a modulation component eliminator according to the first embodiment;
• Fig. 6 shows a configuration of an averaging unit according to the first embodiment;
• Fig. 7 shows a configuration of an estimated signal generator according to the first embodiment;
• Fig. 8 shows another configuration of the estimated signal generator according to the first embodiment;
• Fig. 9 shows a vector diagram of the received signal according to the first embodiment;
• Fig. 10 shows a configuration of a configuration of a fading distortion estimator according to the invention;
• Fig. 11 shows a configuration of a carrier estimator according to the second embodiment of the invention;
• Fig. 12 shows a configuration of an averaging unit according to the sixth embodiment of the invention;
• Fig. 13 shows a configuration of a carrier estimator according to the eighth embodiment of the invention;
• Fig. 14 shows another configuration of the carrier estimator according to the eighth embodiment; Fig. 15 shows a configuration of a receiver of the conventional art;
• Fig. 16 shows a configuration of the fading distortion estimator and compensator according to the conventional art; and
• Fig. 17 shows an example of estimation of fading distortion according to the conventional art.

A demodulator will be explained, which performs coherent detection with compensating an amplitude and a phase of the received signal using a reference signal generated by complex multiplying an average value of the distortion compensated signal to the estimated fading distortion.

Fig. 1 shows a configurational example of a transmitter of the first embodiment performing coherent detection of the received signal using a known signal inserted into transmitted signals.

In the figure, a reference numeral 1 shows a convolutional encoder, 10 shows an interleaving unit, 20 shows a known signal inserting unit, and 30 shows a modulator.

In the following, an operation of the transmitter will be explained.

The convolutional encoder 1 implements convolutional encoding to an information signal and output the code. The interleaving unit 10 arranges the encoded data supplied from the convolutional encoder 1 according to a specific rule. The known signal inserting unit 20 periodically inserts a known signal into the data output from the interleaving unit 10 and output the data. Fig. 2 shows a known signal of R symbol periodically inserted into an information signal (N_R - R) by the known signal inserting unit 20. The modulator 30 modulates the signal supplied from the known signal inserting unit 20 and outputs the modulated signal as a transmitting signal. The transmitting signal is, for example, modulated by QPSK modulating system and transmitted.

Fig. 3 shows a configurational example of the receiver of the first embodiment performing coherent detection of the received signal using the known signal inserted into the transmitted signals.

In the figure, a reference numeral 40 shows a quasi-coherent detecting unit, 50 shows a carrier estimator, 60 shows a data discriminator, 70 shows a di-interleaving unit, and 80 shows a Viterbi decoder.

An operation of the receiver will be explained in the following.

The quasi-coherent detecting unit 40 converts the frequency of the RF/IF (Radio Frequency/Intermediate Frequency) signal by orthogonal detection. The converted signal is further sampled by the A/D (analog to digital) converter to the discrete value, and is output as baseband signals of I channel and Q channel. The carrier estimator 50 obtains the estimated value c_R(t) of the fading distortion c(t) to eliminate the distortion of the signal caused by fading, etc. from the baseband signals of I channel and Q channel. The data discriminator 60 discriminates the transmitted signal from the baseband signals in the I channel and Q channel, which was performed coherent detection and supplied from the carrier estimator 50. The di-interleaving unit 70 rearranges the data, which was arranged by the interleaving unit 20 according to the specific rule, to the original order. The Viterbi decoder 80 decodes the data supplied from the di-interleaving unit 70 using Viterbi algorithm.

Fig. 4 shows a configurational example of the carrier estimator performing coherent detection of the received signal using the known signal inserted into the transmitted signals.

In the figure, a reference numeral 51 shows a fading distortion estimator, 52 shows a fading distortion compensator, 53 shows a modulation component eliminator, 54 shows an averaging unit, 55 shows an estimated signal generator, and 56 shows a signal compensator.

An operation of the carrier estimator 50 will be described below.

The fading distortion estimator 51 estimates the fading distortion of the information signal and operates as described below. The fading distortion compensator 52 compensates the fading distortion using the estimated value of fading distortion obtained by the fading distortion estimator 51, and eliminates the fading distortion from the signal output from the quasi-coherent detecting unit 40. The fading distortion compensator 52 operate, for example, as described in the conventional related art.

Namely, the fading distortion c_R(k-1), c_Rk, c_R(k+1), which were obtained at t = (k-1)T_R, kT_R, (k+1) T_R corresponding to the known signal among the signals output from the quasi-coherent detecting unit 40, are estimated. Using these estimated values, the fading distortion at time t = kT_R + (m/N_R)T_R (N_R: the number of symbols corresponding to the insertion interval of the known signal T_R) is estimated by, for example, the second order Gaussian interpolation and given by the following expression (11).

The fading distortion compensator 52 eliminates the fading distortion from the received signal using the interpolation sequence c_RI(k+m/N_R) supplied from the fading distortion estimator 51. Namely, a transfer function h(k+m/N_R) for compensating the fading distortion is given by the following expression (12).

Using the expression (12), the fading distortion can be compensated as given by the following expression (13) from the signal supplied from the quasi-coherent detecting unit 40.

The modulation component eliminator 53 eliminates the modulation component from the received signal z_RE(k+m/N_R), which was compensated fading distortion and supplied from the fading distortion compensator 52. For example, assuming QPSK modulation signal having four phases of 45°, 135°, 225°, and 315° as signal points, the above elimination corresponds reducing the signal having the first quadrant through the fourth quadrant to only the first quadrant.

Fig. 5 shows a configurational example of the modulation component 53. In the figure, a reference numeral 530 shows an angle converter, 531 shows a residual calculator, 532 shows a π/4 subtractor, and 533 shows a orthogonal coordinates converter.

Next, an operation of the modulation component eliminator 53 will be explained.

The angle converter 530 converts the angles of I data and crossed Q data so as to be phase data. The residual calculator 531 divides the phase data output from the angle converter 530 by 90° and outputs the residue of the division. The π/4 subtractor 532 subtracts 45° from the phase data output from the residual calculator 531. The orthogonal coordinates converter 533 converts the phase data output from the π/4 subtractor 532 into rectangular coordinates. This operation can be represented by the following expression (14), where the phase component of z_RE(k+m/N_R) is assumed to be θ_RE(k+m/N_R) and the signal eliminated by the modulation component eliminator 53 is assumed to be z_RF(k+m/N_R).

In the above expression, Re[·] represents a real part of [·], and Im[·] represents an imaginary part of [·].

The averaging unit 54 performs moving average of the signal of which the modulation component was eliminated by the modulation component eliminator 53 and compensated fading distortion.

Fig. 6 shows a configurational example of the averaging unit 54. In the figure, a reference numeral 540 shows an N-stage shift register, and 541 shows an adder.

In the following, an operation of the averaging unit 54 will be explained.

The N-stage shift register 540 inputs I data and Q data output from the modulation component eliminator 53 every 1 symbol time T_s (= T_R/N_R). The adder 541 performs addition to each symbol of N pieces of data stored in the N-stage shift register 540, and outputs the result of the addition. This operation will be represented by the following expression (15).

The estimated signal generator 55 compounds the estimated value c_RI(k+m/N_R) of the fading distortion obtained by the fading distortion estimator 51 and the mean value z_AV(k+m/N_R) output from the averaging unit 54, and generates a reference signal for estimating the carrier phase and performing coherent detection.

Fig. 7 shows a concept of the estimated signal generator 55. In the figure, a reference numeral 550 shows a complex multiplier. The complex multiplier 550 multiplies the signal output from the averaging unit 54 to the signal output from the fading distortion estimator 51 and outputs the result of the multiplication. This operation equals to the following expression (16) compounding and outputting the reference signal c_EST(k+m/N_R) for coherent detection.

On the other hand, on compounding the signal using only the phase, the estimated signal generator compounds a signal as given by the following expression and outputs the reference signal c_EST(k+m/N_R).

Fig. 8 shows a configurational example of the estimated signal generator 55. In the figure, a reference numerals 551 and 552 show angle converters, 553 shows an adder, and 554 shows an orthogonal coordinates converter.

Next, an operation of the estimated signal generator 55 will be explained.

The angle converter 551 converts the signal output from the averaging unit 54 into the phase data. The angle converter 552 converts the signal output from the fading distortion estimator 51 into the phase data. The adder 553 adds the phase data output from the angle converter 551 and the phase data output from the angle converter 552. The orthogonal coordinates converter 554 converts the phase data output from the adder 553 into the rectangle coordinates, and outputs I data and Q data. This operation equals to the following expression (17) generating the signal and outputting the reference signal c_EST(k+m/N_R) for coherent detection.

The signal compensator 56 compensates the amplitude and the phase of the received signal using the reference signal c_EST(k+m/N_R) output from the estimated signal generator 55 and performs coherent detection. By this operation, the received signal z_RC(k+m/N_R) of the following expression (18) is obtained.

Fig. 9 shows a signal vector diagram of the received signal, where the channel is assumed to be Rician fading channel.

In Rician fading, a signal vector is configured by the direct path signal and multipath component. Further, the thermal noise generated in the receiver is added to the faded signal. The fading distortion estimator 51 mainly estimates the fading distortion based on the signal compounding the direct path signal and the multipath component, and the fading distortion compensator 52 compensates the fading distortion. The averaging unit 54 estimates distortion of the signal mainly caused by the thermal noise, which was not sufficiently compensated by the fading distortion compensator 52. By compounding the estimated values obtained by the fading distortion estimator 51 and by the averaging unit 54, both signal distortion caused by the fading and by the thermal noise can be estimated and compensated to be performed coherent detection.

As has been described, in this embodiment, after the fading distortion is compensated using the known signal, the signal is averaged to detect the estimation error caused by the thermal noise. The obtained estimation error is compounded to the estimated value supplied from the fading distortion estimator to obtain the reference signal to compensate the received signal. The estimation error caused by the thermal noise, which was not sufficiently compensated, can be detected and the degradation of the characteristic caused by the fading and the thermal noise can be reduced. In the first embodiment, the signal is supposed to be modulated by the QPSK modulation method. The embodiment can be applied to PSK having N phases (N=2, 4, 8, etc.). and QAM and so on. In the above explanation, the demodulator of the embodiment includes the convolutional encoder, the Viterbi decoder, the interleaving unit, and di-interleaving unit, however, these elements are not always required to be included in the demodulator.

Fading distortion is estimated not only using the distortion at the location of the known signal, but adding and averaging the signal after weighting the distortion of the previous signal and the subsequent signal. In this way, the distortion can be estimated based on the fading variation with reducing the estimation error due to the thermal noise.

The fading distortion estimator 51b is configured as shown in Fig. 10.

In Fig. 10, a reference numeral 510 shows a known signal extractor, 511 shows a distortion detecting and averaging unit, 512 shows a weighted averaging unit, and 513 shows an interpolator.

In the following, an operation of the fading distortion estimator 51b will be explained.

It is assumed that the known signal having R symbols is periodically inserted with an information signal interval of (N_R -R) symbols as shown in Fig. 2.

Before obtaining the fading distortion estimated value c_RI(k+m/N_R), the fading distortion estimated value c_RB(k) of the known signal having consecutive R symbols is required to be obtained. The known signal extractor 510 extracts the signal corresponding to the known signal. Hereinafter, the extracted known signal having consecutive R symbols is called as a known signal block. The distortion detecting and averaging unit 511 detects the fading distortion in the known signal block output from the known signal extractor 510, and averages the signal. For example, a case of R = 2 symbols will be explained below.

Here, a timing for each symbol is assumed to be as given by the following expression (21).

In this case, T_R shows an insertion interval of the known signal, and N_R shows the number of symbols corresponding to the interval. The timing of the first symbol is assumed to be m = 0 and conditions are assumed to be as given by the following expression (22).$$\mathit{\text{t}}\text{=}{\mathit{\text{kT}}}_{\mathit{\text{R}}}$$

In the below expressions, the estimated value c_RB(k) of the fading distortion in the known signal block is obtained using the estimated values c_R(k), c_R(k+1/N_R) of the fading distortion in each known signal, which is given by the following expression (23).

The weighted averaging unit 512 performs the weighted average of the estimated value c_RB(k) of the fading distortion output from the distortion detecting and averaging unit 511 as shown in the following expression (24).

In the above expression, k represents a variable deciding the number of known signal blocks for weighted averaging. In this way, as well as the estimated value of the fading distortion at the location of known signal block, the estimated values of fading distortion in the previous and subsequent known signal blocks are weighted and added. The added result becomes the estimated value of the fading distortion at that location.

The interpolator 513 estimates the fading distortion C_RI(k+m/N_R) in the information signal using the estimated value c_RBA(k) of the fading distortion at the known signal location output from the weighted averaging unit 512. The conventional way of the interpolation as shown in Fig. 17 can be applied to the above estimation of the fading distortion c_RI(k+m/N_R) in the information signal using the estimated value of fading distortion c_RBA(k). The estimated value c_RI(k+m/N_R) of fading distortion in the information signal can be also obtained by, for example, the second order Gaussian interpolation, and the estimated value can be represented by the following expression (25).

In another way, instead of the second order Gausian interpolation, using the second approximation obtained by the method of least squares, the estimated value c_RI(k+m/N_R) of the fading distortion in the information signal, and the estimated value can be represented by the following expression (26).$${\mathit{\text{c}}}_{\mathit{\text{RI}}}\text{(}{\mathit{\text{t}}}_{\mathit{\text{m}}}\text{) =}{\mathit{\text{c}}}_{\text{0}}\text{+}{\mathit{\text{c}}}_{\text{1}}{\mathit{\text{t}}}_{\mathit{\text{m}}}\text{+}{\mathit{\text{c}}}_{\text{2}}{\mathit{\text{t}}}_{\mathit{\text{m}}}{\text{}}^{\text{\hspace{0.05em}2}}$$$$\mathit{\text{C}}\text{=}{\mathit{\text{R}}}_{\mathit{\text{yy}}}{\text{}}^{\text{\hspace{0.05em}-1}}\text{(3)}{\mathit{\text{R}}}_{\mathit{\text{yu}}}\text{(3)}$$$${\mathit{\text{t}}}_{\mathit{\text{m}}}\text{=}\mathit{\text{k}}\text{+}\frac{\mathit{\text{m}}}{{\mathit{\text{N}}}_{\mathit{\text{r}}}}$$

In the above expression, X^t represents the transposed matrix X.

In another way, instead of the second order Gausian interpolation, using the Wiener interpolation, the estimated value c_RI(k+m/N_R) of the fading distortion in the information signal, and the estimated value can be represented by the following expression (27).

In the expression (27), ρ represents a correlation function, which can be given by a time correlation function as obtained by the following expression (28), considering the fading process.$${\mathit{\text{C}}}_{\mathit{\text{RI}}}\text{(}{\mathit{\text{t}}}_{\mathit{\text{m}}}\text{) =}{\mathit{\text{H}}}^{\text{+}}\mathit{\text{C}}$$$$\mathit{\text{H}}\text{=}{\mathit{\text{R}}}^{\text{-1}}\mathit{\text{W}}$$$${\mathit{\text{r}}}_{\mathit{\text{ij}}}{\text{= ρ}}_{\mathit{\text{ij}}}$$$${\mathit{\text{w}}}_{\mathit{\text{mj}}}{\mathit{\text{= ρ}}}_{\mathit{\text{mj}}}$$$${\mathit{\text{t}}}_{\mathit{\text{m}}}\text{=}\mathit{\text{k}}\text{+}\frac{\mathit{\text{m}}}{{\mathit{\text{N}}}_{\mathit{\text{r}}}}$$

In the above expression, + (dagger) represents a conjugated transposition, a_p represents the known signal, σc² represents dispersion of the amplitude of received signal, J_o(·) represents the zero-order Bessel function of the first kind, f_DT_s represents normalized maximum Doppler frequency, and δ_ij represents Kronecker's delta.

As has been described, the demodulator of the invention averages the estimated value of the fading distortion obtained by the known signal block by adding weighted value to the previous and subsequent signal blocks of the known signal block, and then the interpolation is implemented to estimate the fading distortion. The invention can reduce the estimation error at the known signal location due to the thermal noise, and degradation of the characteristic caused by the thermal noise with following the fading variation.

The value of the weighted coefficient is assumed as given by the expression (24). The weighted coefficient can be set independently.

In the second embodiment, the demodulator having a simpler configuration than the first embodiment, which can reduce the influence of the thermal noise and can be applied to wide range of the characteristic of the channels, will be explained.

In this embodiment, the distortion is not obtained at once, but the distortion of the signal is compensated and then the distortion compensated signal is averaged by each calculation. The reference signal can be obtained as a result of this averaging. The operation from the coherent detection using the reference signal is the same as the first embodiment.

Fig. 11 shows a configurational example of the carrier estimator 50b performing coherent detection of the received signal using the reference signal inserted into the transmitted signals.

In the figure, a reference numeral 51 shows the fading distortion estimator, 52 shows the fading distortion compensator, 53 shows the modulation component eliminator, 54 shows the averaging unit, and 56b shows the signal compensator.

In the following, an operation of the carrier estimator 50b will be explained.

The fading distortion estimator 51 estimates the fading variation in the information signal using the known signal periodically inserted into the received signal. The fading distortion compensator 52 compensates the fading distortion using the fading distortion estimated value estimated by the fading distortion estimator 51, and eliminates the fading distortion from the signal output from the quasi-coherent detector. The modulation component eliminator 53 eliminates the modulation component from the signal output from the fading distortion compensator 52. The averaging unit 54 averages the signal, of which the modulation component was eliminated by the modulation component eliminator 53 and of which the fading distortion was compensated. These operation are processed, for example, in the way as described in the first embodiment.

In the present embodiment, the estimated signal generator is removed. Instead, the signal compensator 56b compensates the signal output from the fading distortion estimator 52 using the signal output from the averaging unit 54. Namely, when the output from the fading distortion estimator 52 is z_RE(k+m/N_R) and the output from the averaging unit 54 is z_AV(k+m/N_R), the received signal z_RC(k+m/N_R) of which the carrier phase was compensated, is given by the following expression (19) when the outputs are compounded to generate a signal including the amplitude.

The signal can be compensated in another way as represented by the following expression (20).

As has been described, the demodulator of the second embodiment first compensates the fading distortion using the known signal, then averages the signal to detect the estimation error due to the thermal noise, and the signal is compensated. The embodiment enables to obtain the demodulator detecting the estimation error due to the thermal noise, which was not sufficiently compensated by the fading distortion compensator, and having a simpler configuration than the first embodiment. Accordingly, the demodulator can reduce the degradation of the performance caused by the fading and the thermal noise.

In the third embodiment, another demodulator will be explained, where the number of pieces of data to be averaged can be varied so as to perform more suitable averaging of the signal to the characteristic of each channel.

In the first and second embodiments, the carrier estimator performs the coherent detection of the received signal using the known signal inserted into the transmitted signals. On the contrary, the averaging unit of the third embodiment is configured as shown in Fig. 12. The other elements of the configuration of the third embodiment are the same as elements of the first embodiment shown in Fig. 4 or the second embodiment shown in Fig. 11.

In Fig. 12, a reference numeral 540 shows an N-stage shift register, 541 shows an adder, 542 shows a tap coefficient setting unit, and 543 shows a multiplier.

An operation of the averaging unit 54b will be explained below.

The N-stage shift register sequentially inputs each signal of I channel and Q channel output from the modulation component eliminator 53. The tap coefficient setting unit 542 outputs N tap coefficients, which can be set varied, corresponding to each shift register. The multiplier 543 multiplies N pieces of data stored in each shift register of the N-stage shift register 542, and N coefficients generated by the tap coefficient setting unit 542, respectively. The adder 541 adds N multiplied results supplied from the multiplier 543 and outputs the result of the addition.

As has been described, the demodulator of the third embodiment averages the signal, of which the modulation component was eliminated, using the N weighted coefficients set by the tap coefficient setting unit 542. The averaging process can be varied according to the channel condition by varying the number of pieces of data and the value of the weighted coefficients. Accordingly, the performance of the receiver has been improved.

In the fourth embodiment, another demodulator will be explained, where the fading distortion is estimated according to the variation of the characteristic of the channel varying in time and averaging parameters are adapted and controlled.

Fig. 13 shows a configuration of the carrier estimator according to the fourth embodiment performing the coherent detection of the received signal using the known signal inserted into the transmitted signals.

In the figure, a reference numeral 57 shows a channel characteristic estimating and controlling unit. The other element of Fig. 13 are the same as the ones of the first embodiment shown in Fig. 4.

An operation of the demodulator shown in Fig. 13 will be explained below.

The channel characteristic estimating and controlling unit 57 estimates the channel using the output signal of the quasi-coherent detecting unit 40, the output from the fading distortion compensator 52 and the output from the signal compensator 56. The weighted coefficient suitable to the present channel is set in the fading distortion estimator 51g using the weighted coefficient control signal, and the averaging process suitable to the present channel is set in the averaging unit 54g by the averaging coefficient control signal. Namely, the weighted coefficient of the weighted averaging unit 512 of the first embodiment shown in Fig. 10 is adaptably determined according to the channel condition. Or, tap coefficient generated by the tap coefficient setting unit 542 of the third embodiment shown in Fig. 12 is adaptably determined.

As has been described, in the demodulator of the fourth embodiment, the channel condition is always estimated in order to set the most suitable control signal to the condition, which enables the subsequent coherent detection to be. performed according to the characteristic variation of the channel. Therefore, the most suitable receiving performance can be obtained on receiving the signal from the fading channel to the nonfading channel.

Fig. 14 shows another configuration of the demodulator of the second embodiment, where the carrier estimator of the fourth embodiment is applied.

In the figure, a reference numeral 57 shows the channel characteristic estimating and controlling unit, and the other elements of Fig. 14 are the same as the elements of the second embodiment shown in Fig. 11.

An operation of the demodulator of Fig. 14 will be explained in the following.

The channel characteristic estimating and controlling unit 57 estimates the channel using the output signal from the quasi-coherent detecting unit 40, the output from the fading distortion compensator 52 and the output from the signal compensator 56. The weighted coefficient suitable to the present channel is set in the fading distortion estimator and the averaging process suitable to the present channel is set in the averaging unit.

As has been described, in the demodulator of the fourth embodiment, the subsequent coherent detection can be performed according the variation of the characteristic of the channel. Therefore, a high receiving performance is obtained on receiving the signal of wide range from the fading channel to the nonfading channel.

As has been described, according to the present invention, after the fading distortion is compensated using the known signal, the signal is processed averaging to detect the estimation error caused by the thermal noise. The obtained estimation error is compounded to the estimated value supplied from the fading distortion estimator to obtain the reference signal to compensate the received signal. The estimation error caused by the thermal noise, which was not sufficiently compensated conventionally by the fading distortion compensator, can be detected and the degradation of the characteristic caused by the fading and the thermal noise can be reduced.

Further, the demodulator of the invention first compensates the fading distortion using the known signal, and then averages the signal to detect the estimation error caused by the thermal noise, and the signal is compensated. The demodulator can detect the estimation error caused by the thermal noise and can reduce the degradation of the characteristics caused by the fading and the thermal noise by a simpler configuration.

Further, the demodulator of the invention averages the estimated value of the fading distortion obtained by the known signal block by adding the weighted value to the previous and subsequent signal blocks of the known signal block. The demodulator can reduce the degradation of the characteristic caused by the thermal noise with adapting the fading variation.

Further, after the output signal of the fading distortion compensator is temporarily discriminated, the modulation component is eliminated by re-modulating the signal using the output from the temporary data discriminator. Since the demodulator of the present invention can eliminate the modulation component more precisely than Nth power law and so on, which reduces the estimation error generated by the averaging unit. Further, the number of stages required for averaging can be decreased, thus the load of the averaging unit can be lessened.

Further, after the data is temporarily discriminated using the output signal from the signal compensator, the modulation component is eliminated by re-modulating the signal using the output from the temporary data discriminator, which enables the demodulator of the present invention to eliminate the modulation component more precisely. Therefore, the estimation error generated by the averaging unit can be reduced. Further, the number of stages required for averaging can be decreased, thus the load of the averaging unit can be lessened.

Further, the demodulator of the present invention averages the signal, of which the modulation component was eliminated, by multiplying tap coefficients. The averaging can be processed according to the channel condition, which improves the performance of the receiver.

Further, in the demodulator of the present invention, the reliability factor is weighted to the data, which was processed by the hard data decision by the data discriminator, using the fading distortion information supplied from the fading distortion estimator. On Viterbi decoding, the Viterbi decoding is implemented by soft decision data according to the reliability factor, so that the error correction ability is improved, and the performance of the receiver has been improved.

Yet further, in the demodulator of the present invention, the channel condition is always estimated in order to set the most suitable control signal to the condition, which enables the subsequent coherent detection can be performed based on the characteristic variation of the channel. Therefore, the most suitable receiving performance can be obtained on receiving the signal from the fading channel to the nonfading channel.

Having thus described several particular embodiments of the present invention, various alterations, modifications, and improvements will readily occur to those skilled in the art. Such alterations, modifications, and improvements are intended to be part of this disclosure, and are intended to be within the scope of the present invention. Accordingly, the foregoing description is by way of example only, and is not intended to be limiting. The present invention is limited only as defined in the following claims and the equivalents thereto.